JingweiNi/test_trip_Qwen3-8B_np_texts_uhead_prm

该数据集包含多个字段，其中包括问题（question）、答案（answer）、输入ID序列（input_ids）、回复（reply）以及多个与模型性能相关的统计字段，如最大序列概率（MaximumSequenceProbability）、平均Token熵（MeanTokenEntropy）和困惑度（Perplexity）。此外，还包括多个与预训练模型相关的字段，如Qwen模型和Universal-PRM模型的相关参数。数据集分为训练集，包含10240个示例，总大小为460,774,017字节。

The dataset includes multiple fields such as questions (question), answers (answer), input ID sequences (input_ids), replies (reply), and several statistics related to model performance, such as Maximum Sequence Probability (MaximumSequenceProbability), Mean Token Entropy (MeanTokenEntropy), and Perplexity. Additionally, it includes fields related to pre-trained models, such as parameters for Qwen models and the Universal-PRM model. The dataset is split into a training set, which contains 10,240 examples and has a total size of 460,774,017 bytes.